The difference is mostly about tone and degree of certainty:

In other words:

Both talk about possibility, but can usually feels stronger and more concrete, while could often feels softer, more speculative, or more sarcastic.

✅ Before we launch, let’s ask what can go wrong.

Here, can sounds practical and immediate. The risks feel real and worth checking now.

✅ If we changed the whole process today, what could go wrong?

Here, could sounds a little more speculative. The speaker is considering a scenario, not necessarily stating that trouble is already close.

✅ Sure, let’s do it without testing anything. What could go wrong?

This is a very common sarcastic use. Could works especially well when the speaker clearly believes many things may go wrong.

✅ A lot can go wrong when nobody checks the numbers.
✅ A lot could go wrong if nobody checked the numbers.

The first sounds more immediate and factual.
The second sounds more conditional or imagined.

The most common mistake is treating can and could as if they always sound exactly the same.

They don’t.

Even when both are grammatically possible, the tone shifts:

Another common mistake is missing how often could is used for humor or dry sarcasm:

✅ What could go wrong?
Very often means:
A lot will probably go wrong.

Use this rule:

A simple memory trick:

can looks the problem in the eye
could keeps one hand on the doorknob

Or even shorter: